mostlyright.core.schemas.earnings_transcript
Earnings-transcript schema (schema.finance.transcript.v1).
One row per transcript segment of an earnings call. Pairs with
schema.finance.fact.v1 (earnings_fact.py) — the fact schema is the
counted-mention ledger; this is the underlying segmented transcript the counter
runs over.
The enum constants (SEGMENT_VALUES / SPEAKER_ROLE_VALUES /
DELIVERY_VALUES) are imported from earnings_fact rather than redefined
here, so the two schemas and the generated TypeScript types cannot drift apart.
delivery vs source. delivery (enum {live, hosted}, default
"live") records the delivery channel only; the per-row source column is
the source identity the validator checks. The live speech-to-text path and the
hosted feed share one source identity ("earnings_call") because both come
from the same public call audio.
Streaming fields. is_final / spoken_at / stream_seq are
optional, so a batch (post-call) row is unchanged: it defaults is_final to
True and carries no stream_seq. is_final denotes speech-to-text segment
finality only (partial vs final transcript text); it never gates settlement
authority.
Classes
Section titled “Classes”| Class | Description |
|---|---|
EarningsTranscriptSchema() | schema.finance.transcript.v1 — one row per transcript segment. |
class mostlyright.core.schemas.earnings_transcript.EarningsTranscriptSchema
Section titled “class mostlyright.core.schemas.earnings_transcript.EarningsTranscriptSchema”Bases: Schema
schema.finance.transcript.v1 — one row per transcript segment.